1 <?xml version='1.0' encoding='utf-8'?>
2 <chapter xmlns="http://docbook.org/ns/docbook"
3 xmlns:xl="http://www.w3.org/1999/xlink" version="5.0" xml:lang="en-US"
4 xml:id="troubleshootingrecovery">
5 <title xml:id="troubleshootingrecovery.title">Troubleshooting
7 <para>This chapter describes what to do if something goes wrong during
8 recovery. It describes:</para>
12 <xref linkend="dbdoclet.50438225_71141" />
17 <xref linkend="dbdoclet.50438225_37365" />
22 <xref linkend="dbdoclet.50438225_12316" />
27 <xref linkend="dbdoclet.lfsckadmin" />
31 <section xml:id="dbdoclet.50438225_71141">
34 <primary>recovery</primary>
35 <secondary>corruption of backing ldiskfs file system</secondary>
36 </indexterm>Recovering from Errors or Corruption on a Backing ldiskfs File
38 <para>When an OSS, MDS, or MGS server crash occurs, it is not necessary to
39 run e2fsck on the file system.
40 <literal>ldiskfs</literal> journaling ensures that the file system remains
41 consistent over a system crash. The backing file systems are never accessed
42 directly from the client, so client crashes are not relevant for server
43 file system consistency.</para>
44 <para>The only time it is REQUIRED that
45 <literal>e2fsck</literal> be run on a device is when an event causes
46 problems that ldiskfs journaling is unable to handle, such as a hardware
47 device failure or I/O error. If the ldiskfs kernel code detects corruption
48 on the disk, it mounts the file system as read-only to prevent further
49 corruption, but still allows read access to the device. This appears as
51 <literal>EROFS</literal>) in the syslogs on the server, e.g.:</para>
52 <screen>Dec 29 14:11:32 mookie kernel: LDISKFS-fs error (device sdz):
53 ldiskfs_lookup: unlinked inode 5384166 in dir #145170469
54 Dec 29 14:11:32 mookie kernel: Remounting filesystem read-only </screen>
55 <para>In such a situation, it is normally required that e2fsck only be run
56 on the bad device before placing the device back into service.</para>
57 <para>In the vast majority of cases, the Lustre software can cope with any
58 inconsistencies found on the disk and between other devices in the file
60 <para>For problem analysis, it is strongly recommended that
61 <literal>e2fsck</literal> be run under a logger, like
62 <literal>script</literal>, to record all
63 of the output and changes that are made to the file system in case this
64 information is needed later.</para>
65 <para>If time permits, it is also a good idea to first run
66 <literal>e2fsck</literal> in non-fixing mode (-n option) to assess the type
67 and extent of damage to the file system. The drawback is that in this mode,
68 <literal>e2fsck</literal> does not recover the file system journal, so there
69 may appear to be file system corruption when none really exists.</para>
70 <para>To address concern about whether corruption is real or only due to
71 the journal not being replayed, you can briefly mount and unmount the
72 <literal>ldiskfs</literal> file system directly on the node with the Lustre
73 file system stopped, using a command similar to:</para>
74 <screen>mount -t ldiskfs /dev/{ostdev} /mnt/ost; umount /mnt/ost</screen>
75 <para>This causes the journal to be recovered.</para>
77 <literal>e2fsck</literal> utility works well when fixing file system
78 corruption (better than similar file system recovery tools and a primary
80 <literal>ldiskfs</literal> was chosen over other file systems). However, it
81 is often useful to identify the type of damage that has occurred so an
82 <literal>ldiskfs</literal> expert can make intelligent decisions about what
83 needs fixing, in place of
84 <literal>e2fsck</literal>.</para>
85 <screen>root# {stop lustre services for this device, if running}
86 root# script /tmp/e2fsck.sda
87 Script started, file is /tmp/e2fsck.sda
88 root# mount -t ldiskfs /dev/sda /mnt/ost
90 root# e2fsck -fn /dev/sda # don't fix file system, just check for corruption
94 root# e2fsck -fp /dev/sda # fix errors with prudent answers (usually <literal>yes</literal>)</screen>
96 <section xml:id="dbdoclet.50438225_37365">
99 <primary>recovery</primary>
100 <secondary>corruption of Lustre file system</secondary>
101 </indexterm>Recovering from Corruption in the Lustre File System</title>
102 <para>In cases where an ldiskfs MDT or OST becomes corrupt, you need to run
103 <literal>e2fsck</literal> to ensure local filesystem consistency, then use
104 <literal>LFSCK</literal> to run a distributed check on the file system to
105 resolve any inconsistencies between the MDTs and OSTs, or among MDTs.</para>
108 <para>Stop the Lustre file system.</para>
112 <literal>e2fsck -f</literal> on the individual MDT/OST that had
113 problems to fix any local file system damage.</para>
114 <para>We recommend running
115 <literal>e2fsck</literal> under script, to create a log of changes made
116 to the file system in case it is needed later. After
117 <literal>e2fsck</literal> is run, bring up the file system, if
118 necessary, to reduce the outage window.</para>
121 <section xml:id="orphan_objects">
124 <primary>recovery</primary>
125 <secondary>orphaned objects</secondary>
126 </indexterm>Working with Orphaned Objects</title>
127 <para>The simplest problem to resolve is that of orphaned objects. When
128 the LFSCK layout check is run, these objects are linked to new files and
130 <literal>.lustre/lost+found/MDT<replaceable>xxxx</replaceable></literal>
131 in the Lustre file system
132 (where MDTxxxx is the index of the MDT on which the orphan was found),
133 where they can be examined and saved or deleted as necessary.</para>
134 <para condition='l27'>With Lustre version 2.7 and later, LFSCK will
135 identify and process orphan objects found on MDTs as well.</para>
138 <section xml:id="dbdoclet.50438225_12316">
141 <primary>recovery</primary>
142 <secondary>unavailable OST</secondary>
143 </indexterm>Recovering from an Unavailable OST</title>
144 <para>One problem encountered in a Lustre file system environment is when
145 an OST becomes unavailable due to a network partition, OSS node crash, etc.
146 When this happens, the OST's clients pause and wait for the OST to become
147 available again, either on the primary OSS or a failover OSS. When the OST
148 comes back online, the Lustre file system starts a recovery process to
149 enable clients to reconnect to the OST. Lustre servers put a limit on the
150 time they will wait in recovery for clients to reconnect.</para>
151 <para>During recovery, clients reconnect and replay their requests
152 serially, in the same order they were done originally. Until a client
153 receives a confirmation that a given transaction has been written to stable
154 storage, the client holds on to the transaction, in case it needs to be
155 replayed. Periodically, a progress message prints to the log, stating
156 how_many/expected clients have reconnected. If the recovery is aborted,
157 this log shows how many clients managed to reconnect. When all clients have
158 completed recovery, or if the recovery timeout is reached, the recovery
159 period ends and the OST resumes normal request processing.</para>
160 <para>If some clients fail to replay their requests during the recovery
161 period, this will not stop the recovery from completing. You may have a
162 situation where the OST recovers, but some clients are not able to
163 participate in recovery (e.g. network problems or client failure), so they
164 are evicted and their requests are not replayed. This would result in any
165 operations on the evicted clients failing, including in-progress writes,
166 which would cause cached writes to be lost. This is a normal outcome; the
167 recovery cannot wait indefinitely, or the file system would be hung any
168 time a client failed. The lost transactions are an unfortunate result of
169 the recovery process.</para>
171 <para>The failure of client recovery does not indicate or lead to
172 filesystem corruption. This is a normal event that is handled by the MDT
173 and OST, and should not result in any inconsistencies between
177 <para>The version-based recovery (VBR) feature enables a failed client to
178 be ''skipped'', so remaining clients can replay their requests, resulting
179 in a more successful recovery from a downed OST. For more information
180 about the VBR feature, see
181 <xref linkend="lustrerecovery" />(Version-based Recovery).</para>
184 <section xml:id="dbdoclet.lfsckadmin">
187 <primary>recovery</primary>
188 <secondary>oiscrub</secondary>
191 <primary>recovery</primary>
192 <secondary>LFSCK</secondary>
193 </indexterm>Checking the file system with LFSCK</title>
194 <para>LFSCK is an administrative tool for checking and repair of the
195 attributes specific to a mounted Lustre file system. It is similar
196 in concept to an offline fsck repair tool for a local filesystem,
197 but LFSCK is implemented to run as part of the Lustre file system
198 while the file system is mounted and in use. This allows consistency
199 checking and repair of Lustre-specific metadata without unnecessary
200 downtime, and can be run on the largest Lustre file systems with
201 minimal impact to normal operations.</para>
202 <para>LFSCK can verify
203 and repair the Object Index (OI) table that is used internally to map
204 Lustre File Identifiers (FIDs) to MDT internal ldiskfs inode numbers, in
205 an internal table called the OI Table. An OI Scrub traverses the OI table
206 and makes corrections where necessary. An OI Scrub is required after
207 restoring from a file-level MDT backup (
208 <xref linkend="dbdoclet.backup_device" />), or in case the OI Table is
209 otherwise corrupted. Later phases of LFSCK will add further checks to the
210 Lustre distributed file system state. LFSCK namespace scanning can verify
211 and repair the directory FID-in-dirent and LinkEA consistency.</para>
212 <para condition='l26'>In Lustre software release 2.6, LFSCK layout scanning
213 can verify and repair MDT-OST file layout inconsistencies. File layout
214 inconsistencies between MDT-objects and OST-objects that are checked and
215 corrected include dangling reference, unreferenced OST-objects, mismatched
216 references and multiple references.</para>
217 <para condition='l27'>In Lustre software release 2.7, LFSCK layout scanning
218 is enhanced to support verify and repair inconsistencies between multiple
220 <para>Control and monitoring of LFSCK is through LFSCK and the
221 <literal>lctl get_param</literal> command. LFSCK supports three types
222 of interface: switch interface, status interface, and adjustment interface.
223 These interfaces are detailed below.</para>
225 <title>LFSCK switch interface</title>
227 <title>Manually Starting LFSCK</title>
229 <title>Description</title>
230 <para>LFSCK can be started after the MDT is mounted using the
231 <literal>lctl lfsck_start</literal> command.</para>
235 <screen>lctl lfsck_start <-M | --device <replaceable>[MDT,OST]_device</replaceable>> \
237 [-c | --create_ostobj <replaceable>on | off</replaceable>] \
238 [-C | --create_mdtobj <replaceable>on | off</replaceable>] \
239 [-d | --delay_create_ostobj <replaceable>on | off</replaceable>] \
240 [-e | --error <replaceable>{continue | abort}</replaceable>] \
242 [-n | --dryrun <replaceable>on | off</replaceable>] \
245 [-s | --speed <replaceable>ops_per_sec_limit</replaceable>] \
246 [-t | --type <replaceable>check_type[,check_type...]</replaceable>] \
247 [-w | --window_size <replaceable>size</replaceable>]</screen>
250 <title>Options</title>
252 <literal>lfsck_start</literal> options are listed and described below.
253 For a complete list of available options, type
254 <literal>lctl lfsck_start -h</literal>.</para>
255 <informaltable frame="all">
257 <colspec colname="c1" colwidth="3*" />
258 <colspec colname="c2" colwidth="7*" />
263 <emphasis role="bold">Option</emphasis>
268 <emphasis role="bold">Description</emphasis>
277 <literal>-M | --device</literal>
281 <para>The MDT or OST target to start LFSCK on.</para>
287 <literal>-A | --all</literal>
291 <para condition='l26'>Start LFSCK on all
292 targets on all servers simultaneously.
293 By default, both layout and namespace
294 consistency checking and repair are started.</para>
300 <literal>-c | --create_ostobj</literal>
304 <para condition='l26'>Create the lost OST-object for
306 <literal>off</literal>(default) or
307 <literal>on</literal>. If not specified, then the default
308 behaviour is to keep the dangling LOV EA there without
309 creating the lost OST-object.</para>
315 <literal>-C | --create_mdtobj</literal>
319 <para condition='l27'>Create the lost MDT-object for
321 <literal>off</literal>(default) or
322 <literal>on</literal>. If not specified, then the default
323 behaviour is to keep the dangling name entry there without
324 creating the lost MDT-object.</para>
330 <literal>-d | --delay_create_ostobj</literal>
334 <para condition='l29'>
335 Delay creating the lost OST-object for dangling LOV EA
336 until the orphan OST-objects are handled.
337 <literal>off</literal>(default) or
338 <literal>on</literal>.
345 <literal>-e | --error</literal>
350 <literal>continue</literal>(default) or
351 <literal>abort</literal>. Specify whether the LFSCK will
352 stop or not if fails to repair something. If it is not
353 specified, the saved value (when resuming from checkpoint)
354 will be used if present. This option cannot be changed
355 while LFSCK is running.</para>
361 <literal>-h | --help</literal>
365 <para>Operating help information.</para>
371 <literal>-n | --dryrun</literal>
375 <para>Perform a trial without making any changes.
376 <literal>off</literal>(default) or
377 <literal>on</literal>.</para>
383 <literal>-o | --orphan</literal>
387 <para condition='l26'>Repair orphan OST-objects for layout
394 <literal>-r | --reset</literal>
398 <para>Reset the start position for the object iteration to
399 the beginning for the specified MDT. By default the
400 iterator will resume scanning from the last checkpoint
401 (saved periodically by LFSCK) provided it is
408 <literal>-s | --speed</literal>
412 <para>Set the upper speed limit of LFSCK processing in
413 objects per second. If it is not specified, the saved value
414 (when resuming from checkpoint) or default value of 0 (0 =
415 run as fast as possible) is used. Speed can be adjusted
416 while LFSCK is running with the adjustment
423 <literal>-t | --type</literal>
427 <para>The type of checking/repairing that should be
428 performed. The new LFSCK framework provides a single
429 interface for a variety of system consistency
430 checking/repairing operations including:</para>
431 <para>Without a specified option, the LFSCK component(s)
432 which ran last time and did not finish or the component(s)
433 corresponding to some known system inconsistency, will be
434 started. Anytime the LFSCK is triggered, the OI scrub will
435 run automatically, so there is no need to specify
436 OI_scrub in that case.</para>
437 <para><literal>namespace</literal>: check and repair
438 FID-in-dirent and LinkEA consistency.</para>
439 <para condition='l27'> Lustre-2.7 enhances
440 namespace consistency verification under DNE mode.</para>
441 <para condition='l26'>
442 <literal>layout</literal>: check and repair MDT-OST
443 inconsistency.</para>
449 <literal>-w | --window_size</literal>
453 <para condition='l26'>The window size for the async request
454 pipeline. The LFSCK async request pipeline's input/output
455 may have quite different processing speeds, and there may
456 be too many requests in the pipeline as to cause abnormal
457 memory/network pressure. If not specified, then the default
458 window size for the async request pipeline is 1024.</para>
467 <title>Manually Stopping LFSCK</title>
469 <title>Description</title>
470 <para>To stop LFSCK when the MDT is mounted, use the
471 <literal>lctl lfsck_stop</literal> command.</para>
475 <screen>lctl lfsck_stop <-M | --device <replaceable>[MDT,OST]_device</replaceable>> \
477 [-h | --help]</screen>
480 <title>Options</title>
482 <literal>lfsck_stop</literal> options are listed and described below.
483 For a complete list of available options, type
484 <literal>lctl lfsck_stop -h</literal>.</para>
485 <informaltable frame="all">
487 <colspec colname="c1" colwidth="3*" />
488 <colspec colname="c2" colwidth="7*" />
493 <emphasis role="bold">Option</emphasis>
498 <emphasis role="bold">Description</emphasis>
507 <literal>-M | --device</literal>
511 <para>The MDT or OST target to stop LFSCK on.</para>
517 <literal>-A | --all</literal>
521 <para>Stop LFSCK on all targets on all servers
522 simultaneously.</para>
528 <literal>-h | --help</literal>
532 <para>Operating help information.</para>
541 <section condition="l29">
542 <title>Check the LFSCK global status</title>
544 <title>Description</title>
545 <para>Check the LFSCK global status via a single
546 <literal>lctl lfsck_query</literal> command on the MDS.</para>
550 <screen>lctl lfsck_query <-M | --device <replaceable>MDT_device</replaceable>> \
552 [-t | --type <replaceable>lfsck_type[,lfsck_type...]</replaceable>] \
553 [-w | --wait]</screen>
556 <title>Options</title>
558 <literal>lfsck_query</literal> options are listed and described below.
559 For a complete list of available options, type
560 <literal>lctl lfsck_query -h</literal>.</para>
561 <informaltable frame="all">
563 <colspec colname="c1" colwidth="3*" />
564 <colspec colname="c2" colwidth="7*" />
569 <emphasis role="bold">Option</emphasis>
574 <emphasis role="bold">Description</emphasis>
583 <literal>-M | --device</literal>
587 <para>The device to query for LFSCK status.</para>
593 <literal>-h | --help</literal>
597 <para>Operating help information.</para>
603 <literal>-t | --type</literal>
607 <para>The LFSCK type(s) that should be queried,
608 including: layout, namespace.</para>
614 <literal>-w | --wait</literal>
618 <para>will wait if the LFSCK is in scanning.</para>
627 <title>LFSCK status interface</title>
629 <title>LFSCK status of OI Scrub via
630 <literal>procfs</literal></title>
632 <title>Description</title>
633 <para>For each LFSCK component there is a dedicated procfs interface
634 to trace the corresponding LFSCK component status. For OI Scrub, the
635 interface is the OSD layer procfs interface, named
636 <literal>oi_scrub</literal>. To display OI Scrub status, the standard
637 <literal>lctl get_param</literal> command is used as shown in the
642 <screen>lctl get_param -n osd-ldiskfs.<replaceable>FSNAME</replaceable>-[<replaceable>MDT_target|OST_target</replaceable>].oi_scrub</screen>
645 <title>Output</title>
646 <informaltable frame="all">
648 <colspec colname="c1" colwidth="3*" />
649 <colspec colname="c2" colwidth="7*" />
654 <emphasis role="bold">Information</emphasis>
659 <emphasis role="bold">Detail</emphasis>
667 <para>General Information</para>
672 <para>Name: OI_scrub.</para>
675 <para>OI scrub magic id (an identifier unique to OI
679 <para>OI files count.</para>
682 <para>Status: one of the status -
683 <literal>init</literal>,
684 <literal>scanning</literal>,
685 <literal>completed</literal>,
686 <literal>failed</literal>,
687 <literal>stopped</literal>,
688 <literal>paused</literal>, or
689 <literal>crashed</literal>.</para>
692 <para>Flags: including -
693 <literal>recreated</literal>(OI file(s) is/are
695 <literal>inconsistent</literal>(restored from
697 <literal>auto</literal>(triggered by non-UI mechanism),
699 <literal>upgrade</literal>(from Lustre software release
700 1.8 IGIF format.)</para>
703 <para>Parameters: OI scrub parameters, like
704 <literal>failout</literal>.</para>
707 <para>Time Since Last Completed.</para>
710 <para>Time Since Latest Start.</para>
713 <para>Time Since Last Checkpoint.</para>
716 <para>Latest Start Position: the position for the
717 latest scrub started from.</para>
720 <para>Last Checkpoint Position.</para>
723 <para>First Failure Position: the position for the
724 first object to be repaired.</para>
727 <para>Current Position.</para>
734 <para>Statistics</para>
740 <literal>Checked</literal> total number of objects
745 <literal>Updated</literal> total number of objects
750 <literal>Failed</literal> total number of objects that
751 failed to be repaired.</para>
755 <literal>No Scrub</literal> total number of objects
757 <literal>LDISKFS_STATE_LUSTRE_NOSCRUB and
758 skipped</literal>.</para>
762 <literal>IGIF</literal> total number of objects IGIF
767 <literal>Prior Updated</literal> how many objects have
768 been repaired which are triggered by parallel
773 <literal>Success Count</literal> total number of
774 completed OI_scrub runs on the target.</para>
778 <literal>Run Time</literal> how long the scrub has run,
779 tally from the time of scanning from the beginning of
780 the specified MDT target, not include the
781 paused/failure time among checkpoints.</para>
785 <literal>Average Speed</literal> calculated by dividing
786 <literal>Checked</literal> by
787 <literal>run_time</literal>.</para>
791 <literal>Real-Time Speed</literal> the speed since last
792 checkpoint if the OI_scrub is running.</para>
796 <literal>Scanned</literal> total number of objects under
797 /lost+found that have been scanned.</para>
801 <literal>Repaired</literal> total number of objects
802 under /lost+found that have been recovered.</para>
806 <literal>Failed</literal> total number of objects under
807 /lost+found failed to be scanned or failed to be
819 <title>LFSCK status of namespace via
820 <literal>procfs</literal></title>
822 <title>Description</title>
824 <literal>namespace</literal> component is responsible for checks
825 described in <xref linkend="dbdoclet.lfsckadmin" />. The
826 <literal>procfs</literal> interface for this component is in the
828 <literal>lfsck_namespace</literal>. To show the status of this
830 <literal>lctl get_param</literal> should be used as described in the
832 <para>The LFSCK namespace status output refers to phase 1 and phase 2.
833 Phase 1 is when the LFSCK main engine, which runs on each MDT,
834 linearly scans its local device, guaranteeing that all local objects
835 are checked. However, there are certain cases in which LFSCK cannot
836 know whether an object is consistent or cannot repair an inconsistency
837 until the phase 1 scanning is completed. During phase 2 of the
838 namespace check, objects with multiple hard-links, objects with remote
839 parents, and other objects which couldn't be verified during phase 1
840 will be checked.</para>
844 <screen>lctl get_param -n mdd. <replaceable>FSNAME</replaceable>-<replaceable>MDT_target</replaceable>.lfsck_namespace</screen>
847 <title>Output</title>
848 <informaltable frame="all">
850 <colspec colname="c1" colwidth="3*" />
851 <colspec colname="c2" colwidth="7*" />
856 <emphasis role="bold">Information</emphasis>
861 <emphasis role="bold">Detail</emphasis>
869 <para>General Information</para>
875 <literal>lfsck_namespace</literal></para>
878 <para>LFSCK namespace magic.</para>
881 <para>LFSCK namespace version..</para>
884 <para>Status: one of the status -
885 <literal>init</literal>,
886 <literal>scanning-phase1</literal>,
887 <literal>scanning-phase2</literal>,
888 <literal>completed</literal>,
889 <literal>failed</literal>,
890 <literal>stopped</literal>,
891 <literal>paused</literal>,
892 <literal>partial</literal>,
893 <literal>co-failed</literal>,
894 <literal>co-stopped</literal> or
895 <literal>co-paused</literal>.</para>
898 <para>Flags: including -
899 <literal>scanned-once</literal>(the first cycle
900 scanning has been completed),
901 <literal>inconsistent</literal>(one or more
902 inconsistent FID-in-dirent or LinkEA entries that have
904 <literal>upgrade</literal>(from Lustre software release
905 1.8 IGIF format.)</para>
908 <para>Parameters: including
909 <literal>dryrun</literal>,
910 <literal>all_targets</literal>,
911 <literal>failout</literal>,
912 <literal>broadcast</literal>,
913 <literal>orphan</literal>,
914 <literal>create_ostobj</literal> and
915 <literal>create_mdtobj</literal>.</para>
918 <para>Time Since Last Completed.</para>
921 <para>Time Since Latest Start.</para>
924 <para>Time Since Last Checkpoint.</para>
927 <para>Latest Start Position: the position the checking
928 began most recently.</para>
931 <para>Last Checkpoint Position.</para>
934 <para>First Failure Position: the position for the
935 first object to be repaired.</para>
938 <para>Current Position.</para>
945 <para>Statistics</para>
951 <literal>Checked Phase1</literal> total number of
952 objects scanned during
953 <literal>scanning-phase1</literal>.</para>
957 <literal>Checked Phase2</literal> total number of
958 objects scanned during
959 <literal>scanning-phase2</literal>.</para>
963 <literal>Updated Phase1</literal> total number of
964 objects repaired during
965 <literal>scanning-phase1</literal>.</para>
969 <literal>Updated Phase2</literal> total number of
970 objects repaired during
971 <literal>scanning-phase2</literal>.</para>
975 <literal>Failed Phase1</literal> total number of objets
976 that failed to be repaired during
977 <literal>scanning-phase1</literal>.</para>
981 <literal>Failed Phase2</literal> total number of objets
982 that failed to be repaired during
983 <literal>scanning-phase2</literal>.</para>
987 <literal>directories</literal> total number of
988 directories scanned.</para>
992 <literal>multiple_linked_checked</literal> total number
993 of multiple-linked objects that have been
998 <literal>dirent_repaired</literal> total number of
999 FID-in-dirent entries that have been repaired.</para>
1003 <literal>linkea_repaired</literal> total number of
1004 linkEA entries that have been repaired.</para>
1008 <literal>unknown_inconsistency</literal> total number of
1009 undefined inconsistencies found in
1010 scanning-phase2.</para>
1014 <literal>unmatched_pairs_repaired</literal> total number
1015 of unmatched pairs that have been repaired.</para>
1019 <literal>dangling_repaired</literal> total number of
1020 dangling name entries that have been
1021 found/repaired.</para>
1025 <literal>multi_referenced_repaired</literal> total
1026 number of multiple referenced name entries that have
1027 been found/repaired.</para>
1031 <literal>bad_file_type_repaired</literal> total number
1032 of name entries with bad file type that have been
1037 <literal>lost_dirent_repaired</literal> total number of
1038 lost name entries that have been re-inserted.</para>
1042 <literal>striped_dirs_scanned</literal> total number of
1043 striped directories (master) that have been
1048 <literal>striped_dirs_repaired</literal> total number of
1049 striped directories (master) that have been
1054 <literal>striped_dirs_failed</literal> total number of
1055 striped directories (master) that have failed to be
1060 <literal>striped_dirs_disabled</literal> total number of
1061 striped directories (master) that have been
1066 <literal>striped_dirs_skipped</literal> total number of
1067 striped directories (master) that have been skipped
1068 (for shards verification) because of lost master LMV
1073 <literal>striped_shards_scanned</literal> total number
1074 of striped directory shards (slave) that have been
1079 <literal>striped_shards_repaired</literal> total number
1080 of striped directory shards (slave) that have been
1085 <literal>striped_shards_failed</literal> total number of
1086 striped directory shards (slave) that have failed to be
1091 <literal>striped_shards_skipped</literal> total number
1092 of striped directory shards (slave) that have been
1093 skipped (for name hash verification) because LFSCK does
1094 not know whether the slave LMV EA is valid or
1099 <literal>name_hash_repaired</literal> total number of
1100 name entries under striped directory with bad name hash
1101 that have been repaired.</para>
1105 <literal>nlinks_repaired</literal> total number of
1106 objects with nlink fixed.</para>
1110 <literal>mul_linked_repaired</literal> total number of
1111 multiple-linked objects that have been repaired.</para>
1115 <literal>local_lost_found_scanned</literal> total number
1116 of objects under /lost+found that have been
1121 <literal>local_lost_found_moved</literal> total number
1122 of objects under /lost+found that have been moved to
1123 namespace visible directory.</para>
1127 <literal>local_lost_found_skipped</literal> total number
1128 of objects under /lost+found that have been
1133 <literal>local_lost_found_failed</literal> total number
1134 of objects under /lost+found that have failed to be
1139 <literal>Success Count</literal> the total number of
1140 completed LFSCK runs on the target.</para>
1144 <literal>Run Time Phase1</literal> the duration of the
1146 <literal>scanning-phase1</literal>. Excluding the time
1147 spent paused between checkpoints.</para>
1151 <literal>Run Time Phase2</literal> the duration of the
1153 <literal>scanning-phase2</literal>. Excluding the time
1154 spent paused between checkpoints.</para>
1158 <literal>Average Speed Phase1</literal> calculated by
1160 <literal>checked_phase1</literal> by
1161 <literal>run_time_phase1</literal>.</para>
1165 <literal>Average Speed Phase2</literal> calculated by
1167 <literal>checked_phase2</literal> by
1168 <literal>run_time_phase1</literal>.</para>
1172 <literal>Real-Time Speed Phase1</literal> the speed
1173 since the last checkpoint if the LFSCK is running
1174 <literal>scanning-phase1</literal>.</para>
1178 <literal>Real-Time Speed Phase2</literal> the speed
1179 since the last checkpoint if the LFSCK is running
1180 <literal>scanning-phase2</literal>.</para>
1190 <section condition='l26'>
1191 <title>LFSCK status of layout via
1192 <literal>procfs</literal></title>
1194 <title>Description</title>
1196 <literal>layout</literal> component is responsible for checking and
1197 repairing MDT-OST inconsistency. The
1198 <literal>procfs</literal> interface for this component is in the MDD
1200 <literal>lfsck_layout</literal>, and in the OBD layer, named
1201 <literal>lfsck_layout</literal>. To show the status of this component
1202 <literal>lctl get_param</literal> should be used as described in the
1204 <para>The LFSCK layout status output refers to phase 1 and phase 2.
1205 Phase 1 is when the LFSCK main engine, which runs on each MDT/OST,
1206 linearly scans its local device, guaranteeing that all local objects
1207 are checked. During phase 1 of layout LFSCK, the OST-objects which are
1208 not referenced by any MDT-object are recorded in a bitmap. During
1209 phase 2 of the layout check, the OST-objects in the bitmap will be
1210 re-scanned to check whether they are really orphan objects.</para>
1213 <title>Usage</title>
1214 <screen>lctl get_param -n mdd.
1215 <replaceable>FSNAME</replaceable>-
1216 <replaceable>MDT_target</replaceable>.lfsck_layout
1217 lctl get_param -n obdfilter.
1218 <replaceable>FSNAME</replaceable>-
1219 <replaceable>OST_target</replaceable>.lfsck_layout</screen>
1222 <title>Output</title>
1223 <informaltable frame="all">
1225 <colspec colname="c1" colwidth="3*" />
1226 <colspec colname="c2" colwidth="7*" />
1231 <emphasis role="bold">Information</emphasis>
1236 <emphasis role="bold">Detail</emphasis>
1244 <para>General Information</para>
1250 <literal>lfsck_layout</literal></para>
1253 <para>LFSCK namespace magic.</para>
1256 <para>LFSCK namespace version..</para>
1259 <para>Status: one of the status -
1260 <literal>init</literal>,
1261 <literal>scanning-phase1</literal>,
1262 <literal>scanning-phase2</literal>,
1263 <literal>completed</literal>,
1264 <literal>failed</literal>,
1265 <literal>stopped</literal>,
1266 <literal>paused</literal>,
1267 <literal>crashed</literal>,
1268 <literal>partial</literal>,
1269 <literal>co-failed</literal>,
1270 <literal>co-stopped</literal>, or
1271 <literal>co-paused</literal>.</para>
1274 <para>Flags: including -
1275 <literal>scanned-once</literal>(the first cycle
1276 scanning has been completed),
1277 <literal>inconsistent</literal>(one or more MDT-OST
1278 inconsistencies have been discovered),
1279 <literal>incomplete</literal>(some MDT or OST did not
1280 participate in the LFSCK or failed to finish the LFSCK)
1282 <literal>crashed_lastid</literal>(the lastid files on
1283 the OST crashed and needs to be rebuilt).</para>
1286 <para>Parameters: including
1287 <literal>dryrun</literal>,
1288 <literal>all_targets</literal> and
1289 <literal>failout</literal>.</para>
1292 <para>Time Since Last Completed.</para>
1295 <para>Time Since Latest Start.</para>
1298 <para>Time Since Last Checkpoint.</para>
1301 <para>Latest Start Position: the position the checking
1302 began most recently.</para>
1305 <para>Last Checkpoint Position.</para>
1308 <para>First Failure Position: the position for the
1309 first object to be repaired.</para>
1312 <para>Current Position.</para>
1319 <para>Statistics</para>
1325 <literal>Success Count:</literal> the total number of
1326 completed LFSCK runs on the target.</para>
1330 <literal>Repaired Dangling:</literal> total number of
1331 MDT-objects with dangling reference have been repaired
1332 in the scanning-phase1.</para>
1336 <literal>Repaired Unmatched Pairs</literal> total number
1337 of unmatched MDT and OST-object pairs have been
1338 repaired in the scanning-phase1</para>
1342 <literal>Repaired Multiple Referenced</literal> total
1343 number of OST-objects with multiple reference have been
1344 repaired in the scanning-phase1.</para>
1348 <literal>Repaired Orphan</literal> total number of
1349 orphan OST-objects have been repaired in the
1350 scanning-phase2.</para>
1354 <literal>Repaired Inconsistent Owner</literal> total
1355 number.of OST-objects with incorrect owner information
1356 have been repaired in the scanning-phase1.</para>
1360 <literal>Repaired Others</literal> total number of.other
1361 inconsistency repaired in the scanning phases.</para>
1365 <literal>Skipped</literal> Number of skipped
1370 <literal>Failed Phase1</literal> total number of objects
1371 that failed to be repaired during
1372 <literal>scanning-phase1</literal>.</para>
1376 <literal>Failed Phase2</literal> total number of objects
1377 that failed to be repaired during
1378 <literal>scanning-phase2</literal>.</para>
1382 <literal>Checked Phase1</literal> total number of
1383 objects scanned during
1384 <literal>scanning-phase1</literal>.</para>
1388 <literal>Checked Phase2</literal> total number of
1389 objects scanned during
1390 <literal>scanning-phase2</literal>.</para>
1394 <literal>Run Time Phase1</literal> the duration of the
1396 <literal>scanning-phase1</literal>. Excluding the time
1397 spent paused between checkpoints.</para>
1401 <literal>Run Time Phase2</literal> the duration of the
1403 <literal>scanning-phase2</literal>. Excluding the time
1404 spent paused between checkpoints.</para>
1408 <literal>Average Speed Phase1</literal> calculated by
1410 <literal>checked_phase1</literal> by
1411 <literal>run_time_phase1</literal>.</para>
1415 <literal>Average Speed Phase2</literal> calculated by
1417 <literal>checked_phase2</literal> by
1418 <literal>run_time_phase1</literal>.</para>
1422 <literal>Real-Time Speed Phase1</literal> the speed
1423 since the last checkpoint if the LFSCK is running
1424 <literal>scanning-phase1</literal>.</para>
1428 <literal>Real-Time Speed Phase2</literal> the speed
1429 since the last checkpoint if the LFSCK is running
1430 <literal>scanning-phase2</literal>.</para>
1442 <title>LFSCK adjustment interface</title>
1443 <section condition='l26'>
1444 <title>Rate control</title>
1446 <title>Description</title>
1447 <para>The LFSCK upper speed limit can be changed using
1448 <literal>lctl set_param</literal> as shown in the usage below.</para>
1451 <title>Usage</title>
1452 <screen>lctl set_param mdd.${FSNAME}-${MDT_target}.lfsck_speed_limit=
1453 <replaceable>N</replaceable>
1454 lctl set_param obdfilter.${FSNAME}-${OST_target}.lfsck_speed_limit=
1455 <replaceable>N</replaceable></screen>
1458 <title>Values</title>
1459 <informaltable frame="all">
1461 <colspec colname="c1" colwidth="3*" />
1462 <colspec colname="c2" colwidth="7*" />
1469 <para>No speed limit (run at maximum speed.)</para>
1474 <para>positive integer</para>
1477 <para>Maximum number of objects to scan per second.</para>
1485 <section xml:id="dbdoclet.lfsck_auto_scrub">
1486 <title>Auto scrub</title>
1488 <title>Description</title>
1490 <literal>auto_scrub</literal> parameter controls whether OI scrub will
1491 be triggered when an inconsistency is detected during OI lookup. It
1492 can be set as described in the usage and values sections
1494 <para>There is also a
1495 <literal>noscrub</literal> mount option (see
1496 <xref linkend="mount.lustre" />) which
1497 can be used to disable automatic OI scrub upon detection of a
1498 file-level backup at mount time. If the
1499 <literal>noscrub</literal> mount option is specified,
1500 <literal>auto_scrub</literal> will also be disabled, so OI scrub will
1501 not be triggered when an OI inconsistency is detected. Auto scrub can
1502 be renabled after the mount using the command shown in the usage.
1503 Manually starting LFSCK after mounting provides finer control over
1504 the starting conditions.</para>
1507 <title>Usage</title>
1508 <screen>lctl set_param osd_ldiskfs.${FSNAME}-${MDT_target}.auto_scrub=<replaceable>N</replaceable></screen>
1510 <replaceable>N</replaceable>is an integer as described below.</para>
1511 <note condition='l25'><para>Lustre software 2.5 and later supports
1512 <literal>-P</literal> option that makes the
1513 <literal>set_param</literal> permanent.</para></note>
1516 <title>Values</title>
1517 <informaltable frame="all">
1519 <colspec colname="c1" colwidth="3*" />
1520 <colspec colname="c2" colwidth="7*" />
1527 <para>Do not start OI Scrub automatically.</para>
1532 <para>positive integer</para>
1535 <para>Automatically start OI Scrub if inconsistency is
1536 detected during OI lookup.</para>
1548 vim:expandtab:shiftwidth=2:tabstop=8: